Segmentation and tracking of mesoscale eddies in numeric ocean models

Abstract

An adequate understanding of climate variability and the eventual prediction of climate change are among the most urgent and far-reaching efforts of the scientific community. Measurements done to-date have suggested that the mesoscale eddies and mesoscale features play a strong role in carrying heat poleward. Oceanographic visualizations may play a key role in unraveling these mysteries. In this paper -we propose techniques for the automated detection, segmentation and tracking of these eddies. A structuring element based on velocity data is designed to detect centers of eddies. Best fitting ellipses/circular donuts are determined to outline and segment out the region of swirling currents around the detected centers. Segmented images from consecutive days of data are used to track eddies temporally. Small-motion assumption is used to track the full trajectory of the eddies. Visualizations resulting from this project provides oceanographers an invaluable tool to assess mesoscale eddies and the Lagrangian characteristics of this mesoscale oceanic feature.